基于无人机遥感图像纹理与植被指数的土壤含盐量反演  

Inversion of Soil Salt Content Based on Texture Feature and Vegetation Index of UAV Remote Sensing Images

作　　者：向友珍[1,2] 李汪洋 台翔 安嘉琪 王辛 陈俊英[1,2] XIANG Youzhen;LI Wangyang;TAI Xiang;AN Jiaqi;WANG Xin;CHEN Junying(Key Laboratory of Agricultural Soil and Water Engineering in Arid and Semiarid Areas,Ministry of Education,Northwest A&F University,Yangling,Shaanxi 712100,China;College of Water Resources and Architectural Engineering,Northwest A&F University,Yangling,Shaanxi 712100,China)

机构地区：[1]西北农林科技大学旱区农业水土工程教育部重点实验室,陕西杨凌712100 [2]西北农林科技大学水利与建筑工程学院,陕西杨凌712100

出　　处：《农业机械学报》2023年第8期201-210,共10页Transactions of the Chinese Society for Agricultural Machinery

基　　金：国家自然科学基金面上项目(52279047)。

摘　　要：基于无人机遥感技术获取农田土壤盐分信息为盐渍化治理提供了快速、准确、可靠的理论依据。本文在内蒙古河套灌区沙壕渠灌域试验地上采集了取样点0~20 cm的土壤含盐量,并使用M600型六旋翼无人机平台搭载Micro-MCA多光谱相机采集图像。利用Otsu算法对多光谱图像进行图像分类(土壤背景和植被冠层),基于分类结果分别提取剔除土壤背景前后的光谱指数和图像纹理特征,采用支持向量机(SVM)和极限学习机(ELM)构建土壤含盐量监测模型,其4种建模策略分别为:未剔除土壤背景的光谱指数(策略1)、剔除土壤背景后的光谱指数(策略2)、未剔除土壤背景的光谱指数+图像纹理特征(策略3)、剔除土壤背景的光谱指数+图像纹理特征(策略4),通过比较4种建模策略的模型精度以筛选出最优变量组合。结果表明:策略3、4所计算出的土壤含盐量反演精度高于策略1、2,策略1~4验证集决定系数R_(v)^(2)分别为0.614、0.640、0.657、0.681,因此利用图像纹理特征+植被指数对提高土壤含盐量的反演精度有重要意义。对比策略3、4,图像纹理特征+植被指数受到土壤背景的影响,策略4精度低于策略3精度,其R_(v)^(2)分别为0.614、0.657;各变量处理的最优模型均为ELM模型,建模集R_(v)^(2)分别为0.625、0.644、0.618、0.683,标准均方根误差分别为0.152、0.134、0.206、0.155。相比于SVM模型,ELM模型提高了土壤含盐量的反演精度。The acquisition of farmland soil salt information based on UAV remote sensing technology provides a rapid,accurate and reliable theoretical basis for salinization management.The soil salt content of 0~20 cm from the sampling point was collected on the test ground of Shahao canal irrigation field in Hetao Irrigation District,Inner Mongolia,and the images were collected by M600 hexarotor UAV platform equipped with Micro-MCA multispectral camera.Otsu algorithm was used to classify the multi-spectral images(soil background and vegetation canopy).Based on the classification results,the spectral index and image texture features before and after removing the soil background were extracted respectively.The soil salt content monitoring model was constructed by support vector machine(SVM) and extreme learning machine(ELM).The four modeling strategies were as follows:spectral index of the soil background was not removed(strategy 1);spectral index of the soil background was removed(strategy 2);spectral index of the soil background was not removed + image texture features(strategy 3);spectral index of the soil background was removed + image texture features(strategy 4).The optimal variable combination was selected by comparing the model accuracy of the four modeling strategies.The results showed that the inversion accuracy of soil salt content calculated by strategy 3 and strategy 4 was higher than that of strategy 1 and strategy 2,and their validation sets R_(v)^(2) were 0.614,0.640,0.657 and 0.681,respectively.Therefore,it was of great significance to use image texture feature and vegetation index to improve the inversion accuracy of soil salt content.By comparing strategies 3 and 4,the image texture feature + vegetation index was affected by soil background.The accuracy of the strategy 4 was lower than that of the strategy 3,whose R_(v)^(2) was 0.614 and 0.657,respectively.The optimal model for each variable processing was ELM model,and the modeling sets R_(v)^(2) were 0.625,0.644,0.618,0.683,and the standard root mean squ

关 键 词：土壤含盐量 无人机遥感 多光谱 图像纹理特征 植被指数 全子集筛选 

分 类 号：S127[农业科学—农业基础科学] TP79[自动化与计算机技术—检测技术与自动化装置]